Remote Actuation Mechanism for MR-compatible Manipulator Using Leverage and Parallelogram

Size: px

Start display at page:

Download "Remote Actuation Mechanism for MR-compatible Manipulator Using Leverage and Parallelogram"

Dale Payne
5 years ago
Views:

1 Remote Actuation Mechanism for MR-compatible Manipulator Using Leverage and Parallelogram Workspace Analysis, Workspace Control, and Stiffness Evaluation Yoshihiko Koseki, Noriho Koyachi, Tatsuo Arai and Kiyoyuki Chinzei National Institute of Advanced Industrial Science and Technology (AIST) Namiki, Tsukuba, Ibaraki , Japan {koseki-y, k.chinzei, Graduate School of Engineering Science, Osaka University 1-3 Machikaneyama-cho, Toyonaka, Osaka , Japan Abstract In this paper, a novel mechanism of remote actuation was proposed and its mechanical problems were discussed. This mechanism consists of leverage and parallelogram mechanism and transmits 3 d.o.f translational and 3 d.o.f rotational motions from the input to output. Such a remote actuation is significantly useful for robotic assist around MRI where the strong and precise magnet is used for imaging, because mutual effect between MRI and robot are inverse proportional to the distance. In this paper, the basic ideas of mechanism were introduced firstly. The discussions of its stiffness and workspace conclude their trade-off. Our new idea of workspace control for mechanical safety is preliminary discussed. I. INTRODUCTION A. Intraoperative MRI in Keyhole Surgery Less invasive to normal tissue while surgical procedures, results in less pain, scar and recovery time. The minimally invasive surgeries have been developed recently. These kinds of surgery are often called keyhole surgery because of its narrow vision availability. To compensate the narrow view, tomography like MRI (Magnetic Resonance Imaging) and X-ray CT (X-ray Computer Tomography) has been used intraoperatively in many kinds of minimally invasive surgeries [1], [2], [3], [4]. MRI is superior to X-ray CT in terms of good soft tissue contrast, lack of ionizing radiation, and functional imaging. In conventional MRI, the patient is so closely enclosed by magnets that there is no space for the surgeon(s) to perform procedures. However, MRI which has a relatively wide opening has been developed and spread recently. This kind of MRI is called Open MRI. But the opening is still narrow for sufficient number of surgeons and assistants to stand by the patient. In addition, manual positioning of tools is not as precise as the coordinate information provided by tomography. B. MR-compatible Manipulator A robot can position and manipulate devices and tools quickly and precisely referring to the numerical coordinate frame of tomography [5], [6]. In addition, a robot can work in narrow space with surgeon(s). Therefore some manipulators which are numerically controlled and work inside MRI gantry have been proposed. Masamune has proposed one for stereotactic neurosurgery [7]. Kaiser also has proposed one for stereotactic biopsy and therapy of breast cancer [8]. We also have studied MR compatible mechanics and electronics to develop not only robotic manipulators but also other devices, which can work around MRI [9]. We have developed 5 d.o.f (degrees of freedom) manipulator for brachytherapy of prostate cancer [10]. The technical drawbacks of MRI are that the most conventional materials and devices affect and are affected by the strong and precise magnet of MRI. Steel, normal bearings, and motors cause serious noise and deformation of images and/or don t work normally. A magnetic material like steel is dangerous because it will be attracted strongly. MR-compatibility is necessary for devices used around MRI [11]. The conventional MR-compatible manipulators were mainly made of duralumin, titanium, and engineering plastic and consisted of ultrasonic motors and ceramics bearings. However these materials and parts are inferior in terms of performance, durability, stiffness, and cost. So conventional prototypes have been less powerful and less stiff but even more expensive. C. Research Target Not only exchanging conventional materials and parts with less magnetic one, a remote actuation can improve MR-compatibility because the mutual influences of magnets decrease inverse proportionally to the distance. If actuators can be set away from imaging area, a lot of conventional parts which are superior in terms of cost and

2 Input of Transmitter z Universal Joint 1 Parallelogram Rotating & Sliding Pair Fulcrum Ball Joint Link Lever Universal Joint Wires Input/Output Plate x y Universal Joint 3 Lever Fig. 2. Implementation of parallelogram mechanism x Translational Transmission x, y: Leverage Motion around the Fulcrum z : Sliding Motion along the Lever Fig. 1. z y Universal Joint 2 Output of Transmitter θ x θz θ y Rotational Transmission θ x, θ y : Parallelogram Mechanism θ z : Rotational Motion around the Lever Leverage and parallelogram mechanism performance can be used. In addition, the remote actuation is profitable because the bulky actuators can be installed away from the narrow area around patient. For these reasons, we have studied a remote actuation mechanism using leverage and parallelogram for robotic assist in MR-guided surgery [12]. This mechanism transmits the input s motions to the output. The leverage and slider of the fulcrum allow 3 d.o.f translational motions. The parallelograms and rotation around the lever allow 3 d.o.f rotational motions. This mechanism remains straight always. So it never occupies large space. The fulcrum supports the lever to reduce the deflection and vibration of the long lever. In this paper, the basic ideas of this mechanism were introduced and its merits and demerits were discussed firstly. Secondly, the workspace of our current prototype was numerically analyzed. Thirdly, some numerical examples of our new idea of workspace control were discussed. Fourthly, the relationships between elasticity and lever ratio were tested experimentally. II. LEVERAGE AND PARALLELOGRAM MECHANISM A. Mechanism LPM (Leverage and Parallelogram mechanism) can transmit 3 d.o.f translational and 3 d.o.f rotational motions from input to output. This mechanism itself does not include any actuators. Fig. 1 explains the structure and functions of LPM. The input of transmitter is the one end of lever and is attached to a conventional manipulator. The output is the other end of lever and is allocated into the workspace around MR gantry. The lever is jointed with input and output of transmitter via universal joint 1, 2. These universal joints constrain 3 d.o.f translational motion and rotational motion around the lever, but allow rotational motions vertical to lever. The lever is supported in the middle by universal joint 3 via rotating & sliding pair. The universal joint 3 is fixed on the base and works as a fulcrum of the lever. These parts enable leverage motion and transmit translational motions in x- and y-axis to the output oppositely. The motions along and around the lever are allowed by the rotating & sliding pair. Some links joint the input and output plates via ball joints. The pairs of link and lever form parallelogram. These parallelogram mechanisms always keep the input and output plates in parallel and transmit rotational motion around x- and y-axes. If the links can endure both expansion and contraction, 2 links are necessary and sufficient. If the links can endure only expansion, 3 links are necessary and sufficient. In Fig. 1, the links are set outside of the lever. Fig. 2 shows actual implementation of parallelogram mechanism. The links can be set inside hollow cylinder to avoid the links conflicts with the universal joint 3. In Fig. 2, wires were used instead of links because a wire doesn t need ball joint. 4 wires were installed for symmetry. The intermediate part of wire was replaced with thin shaft to reduce wire s extension. B. Features a) Remote Actuation: An actuator causes a magnetic and electrical noise and its effect on MRI is inverse proportional to the distance. This mechanism can transmit actuation from the distance and enables the actuators to be set in the distance. Even if the actuator is perfectly MRcompatible like hydraulic actuator, the remote actuation keeps the bulky driving members away and saves the space around the imaging area for the surgeon. It s important because even Open MRI has narrow opening. So, this feature improves the MR-compatibility in terms of noise reduction and space saving. b) Surgeon Friendly: A serial link manipulator, which is widely used in industry, has potential danger of conflict between elbow joint and surgeon. Because an articular manipulator causes the unforecastable motion of elbow. LPM doesn t have such danger, because it s always straight. On the other hand, a parallel mechanism 653

3 occupies a large space, because many links are jointed to the endplate in parallel. c) Workspace Control: The wider workspace expands the potentiality of danger because everything in the workspace might be damaged in case of controller crash or human errors. So, the manipulator s motion should be restricted mechanically to secure safety in an emergency. As for LPM, the output workspace can be restrained by the input workspace. For example, if a barrier is set to the left in the input workspace, the end-effector in the output workspace cannot move to the right. The barrier in the input workspace can be scapegoat of patient as far as the barrier is sufficiently strong. The mechanical workspace might be controlled by changing the position and shape of the barrier. So this mechanism might be flexible to different types of operation to some extents. d) Stiffness Reduction: Because LPM is attached to a manipulator in serial, the backlash and deformation LPM are added to that of the manipulator. So LPM decreases the stiffness. e) Workspace Reduction: The inclination of the output is limited by the limit angle of pivots, universal joints, and by conflicts between the lever and rods. The translational displacement along the lever is restricted by the range of sliding pair. The translational workspace changes according to the ratio of the output lever length to input lever length. III. NUMERICAL ANALYSIS AND EXPERIMENTS A. Current Prototype A prototype of LPM manipulator was designed and made to reveal mechanical problems. For this purpose, this prototype was not MR compatible but all components were designed to be exchangeable with MR compatible components. This prototype was designed for so-called doubledoughnut type of Open MRI, GE s Signa SP TM. Fig. 3 shows CG image of the prototype and Signa SP. Signa SP has two magnets with 0.5[m] horizontal opening. Each magnet has an inside diameter of 0.5[m] and outside diameter of 2.2[m]. This type has a horizontally narrow and vertically wide opening, where surgeons can stand by the patient. So the manipulator was arranged above the surgeon and approached to the patient over surgeon s head (See Fig. 3). Signa SP has 0.5[T] horizontal magnetic field. LPM was driven by a fixed-linear parallel mechanism [13]. A fixed-linear parallel mechanism has 6 linear actuators fixed on the base, and has 6 middle-links, which connect between the endplate and actuators in parallel. This type has good property of remote actuation because its actuators always stay in the same positions distant from MR gantry. It has demerit that it has narrow workspace especially in orientation but LPM also Fig. 4. Parallel Mechanism (Fixed Linear Type) Fig. 3. Lever MR Gantries (GE's Double Doughnuts Type) CG view of prototype and open MRI Parallel Mechanism (Fixed Linear Type) Fulcrum End-Effector prototype of leverage and parallelogram mechanism manipulator has narrow workspace in orientation. So this demerit is not bottleneck of workspace. The kinematic parameters were decided for the manipulator to take any position and orientation within ±100[mm] in x-, y-, z-axes, and ±30[deg.] around x-, y-, z-axes. Fig. 4 show our first prototype. The kinematic parameters are shown in Fig. 5. The origins of input was located at the crossing points of universal joint and that of output was located 50[mm] away from the crossing points of universal joints along the lever. 654

4 Linear Actuator 550[mm] 64[mm] Proc. of IEEE International Conference on Robotics and Automation(ICRA), pp , [deg] 71.5[mm] Endplate Middle Link 330[mm] 17[deg] Fulcrum (0,437,750)[mm] z 1168[mm] 215[mm] Center of Base (0,600,600)[mm] 50[mm] y Center of Imaging Area (0,0,0)[mm] x Fig. 5. Link parameters of prototype B. Workspace Analysis The workspace of the prototype was analyzed numerically. The workspace was judged by the following conditions. 1) Inverse kinematics of fixed-linear parallel mechanism 2) Jacobian singularity of fixed-linear parallel mechanism 3) Linear actuators strokes 4) Middle links conflict with middle links, linear actuators, and lever 5) Endplate s conflict with linear actuators and ceiling 6) Allowable angle of universal joints of middle links 7) Stroke of sliding pair 8) Allowable angle of universal joint 1, 2, and 3 In Fig. 6, the workspaces of the input and output are shown on the same coordinate system. The input workspace is shown in upper right and the output workspace is shown in lower left. The solid model indicates the area where the input/output could move keeping the input/output plate horizontal. The gray scale indicates the minimum of maximum rotations around x-, y-, and z-axis independently. The 1/4 of model is sectioned to show rotational limitation of middle of workspace. The input workspace was horizontally restricted by the conflicts between endplate and linear actuators. Vertically it was restricted by the strokes of the linear actuators and by the ceiling. So the input workspace formed vertically long prism. Because of leverage, the workspace, which Fig. 6. Workspace analysis of prototype is closer to the fulcrum is more amplified in the output workspace. So the output workspace formed a sector. In this workspace analysis, the conflict between lever and MR-magnets were not taken into consideration. So the actual workspace would be smaller. The rotational workspace was mainly restricted by singularity of the parallel mechanism and allowable angle of universal joints (< 60[Deg]). The lever ratio was large because the input area where large rotational motions were allowed had to be expanded to ±100[mm] in the output space. C. Workspace Control In this section, some examples of workspace control were numerically analyzed. The conflicts between the endplate and a flat plate were added to the conditions of workspace analysis. Practically, a flat plate cannot be installed to the input space, because it will interfere with the lever and middle links. A plane linkage like one shown in Fig. 7(f), can move passively and avoid conflict with lever and middle links, but mechanically restrict the overshoot of the endplate. Fig. 7 (a)-(e) shows results of workspace analysis. Fig. 7 (a) and (d) are side view and front view of workspace where no barrier was installed. Fig. 7(b) is the results where the plate barrier restricted the output from entering into the lower side (z < 0) assuming that the patient lies this area. Fig. 7 (c) and (e) show the results where back side (y > 0) and right side (x < 0) were protected. 655

5 (a) No barrier(side view) (b) x-y plane barrier(side view) (c) z-x plane barrier(side view) Endplate Barrier (d) No barrier(front view) (e) y-z plane barrier(front view) (f) Plane links barrier Fig. 7. Examples of workspace control The boundary in the input space was roughly similar to the plane which was line symmetric to the barrier plate. The barrier plates are not indicated in Fig. 7. But they correspond to the cut ends of the input workspace. The workspace which should be restricted changes according to types of operation. So the shape of barrier should be calculated individually. Even if the manipulator crashes into the barrier, the safety of patient is secured as long as the barrier is stiff. However, to protect the manipulator from the crash, some kinds of safety mechanism like collision detection and forced outage are necessary. D. Elasticity Evaluation If the lever ratio is n = l output /l input and f is exerted on the lever vertically, the input manipulator receives n f. In addition, if the input manipulator moves δp vertically, the output of lever is displaced n δp. Totally, when the lever ratio is n and the elasticity of the input manipulator is k, that of the output of lever would be n 2 k. The elasticity of the input manipulator is amplified by leverage proportionally to the square of lever ratio. The small deformation of manipulator becomes large at the end of lever if lever ratio is large. Fig. 8 shows the experimental results of prototype s elasticity in x-axis vs lever ratio (The coordinate system is shown in Fig. 5). Here, elasticity means deformation per load. When the manipulator stayed in the home position (lever ratio=2.3), the elasticity of fixed linear parallel mechanism was 0.017[mm/N]. A square polynomial approximation is also plotted. The results show that the elasticity increased quadratically according to lever ratio. The elasticity of the input (0.017[mm/N]) was smaller than the efficiency of square term of the approximation (0.036[mm/N]). That indicates that the elasticity increased because of not only input manipulator s elasticity expanded by leverage but also lever s elasticity. The lever of this prototype had φ50[mm] 656

6 Kx[mm/N] Fig. 8. Stiffness at lever output Polynomial Approximation y=0.036x*x-0.023x Ratio of Lever Stiffness in x-axis to lever ratio at lever output and input outer diameter, φ46[mm] inner diameter duralumin hollow cylinder. The elasticity of lever also increases according to lever ratio. For any reasons, the lever ratio increased the total elasticity. The offset was caused by elasticity of the parallelograms and basement. IV. CONCLUSIONS In this paper, we have introduced our novel mechanism of a remote actuation using leverage and parallelogram for robotic assist in MR-guided surgery. Secondly, the workspace of our current prototype was numerically analyzed. The results showed that the workspaces were amplified according to the lever ratio. Thirdly, some numerical examples of our new idea of workspace control were discussed. Fourthly, the relationships between elasticity and lever ratio were tested experimentally. The results showed that the elasticity increased proportionally to the square of lever ratio. The workspace analysis and elasticity evaluation revealed the tradeoff between payload, speed, workspace, and stiffness. Based on these mechanical problems and discussions, MR-compatible prototype was designed and is under construction. REFERENCES [1] Interactive Image-Guided Neurosurgery, American Association of Neurological Surgeons (AANS), ISBN [2] R.S. Pergolizzi, et al.: Intra-Operative MR Guidance During Trans-Sphenoidal Pituitary Resection: Preliminary Results, J. of Magnetic Resonance Imaging, Vol. 13, pp , [3] DSJ Choy, et al.: Percutaneous laser nucleolysis of lumbar disc, N Engl. J. Med., 317, pp , [4] C.K. Kuhl, et al.: Interventional breast MR imaging: clinical use of a stereotactic localization and biopsy device, Radiology, Vol. 204, pp , [5] C.W. Burckhardt, et al.: Stereotactic Brain Surgery, IEEE Engineering in Medicine and Biology Magazine, Vol. 14, No. 3, pp , [6] Y.S. Kwoh, et al.: A Robot with Improved Absolute Positioning Accuracy for CT Guided Stereotactic Brain Surgery, IEEE tran. on Biomedical Engineering, Vol. 35, No. 2, pp , [7] K. Masamune, et al.: Development of an MRI-compatible needle insertion manipulator for stereotactic neurosurgery, Journal of Image Guided Surgery, Vol. 1, No. 4, pp , [8] W.A. Kaiser, et al.: Robotic system for Biopsy and Therapy in a high-field whole-body Magnetic-Resonance-Tomograph. Proc. Intl. Soc. Mag. Reson. Med., Vol. 8, p. 411, [9] K. Chinzei, et al.: MR Compatibility of Mechatronic Devices: Design Criteria, Proc. of MICCAI 99, pp , [10] K. Chinzei, et al.: MR Compatible Surgical Assist Robot System Integration and Preliminary Feasibility Study, Proc. of MICCAI 2000, pp , [11] F.A Jolesz, et al.: Compatible instrumentation for intraoperative MRI: Expanding resources, J. of Magnetic Resonance Imaging, Vol. 8, No. 1, pp. 8-11, [12] Y. Koseki, et al.: Robotic Assist for MR-Guided Surgery Using Leverage and Parallelepiped Mechanism, Proc. of MICCAI 2000, pp , [13] T. Arai, et al.: Development of a new parallel manipulator with fixed linear actuator, Proc. of Japan/USA Symposium on Flexible Automation, pp ,

Precise Evaluation of Positioning Repeatability of MR-Compatible Manipulator Inside MRI

Precise Evaluation of Positioning Repeatability of MR-Compatible Manipulator Inside MRI Yoshihiko Koseki 1, Ron Kikinis 2, Ferenc A. Jolesz 2, and Kiyoyuki Chinzei 1 1 National Institute of Advanced Industrial